Using Storage DeDupe for AlwaysOn Availability Groups

  • I have a Sr. Sysadmin here who wants to turn on storage dedupe for the three nodes that will be host AG's because he really wants space savings.  The setup is that I actually have 3 sets of 3 nodes.  Each set of three nodes will be hosting databases from what is now a single stand alone instance of either SQL 2008 or SQL 2008R2.  We're apparently running low on flash storage in the SAN and he wants to dedupe the databases themselves. he and I researched this about a year ago in a different scenario and I convinced him it was a bad idea.  I can't find the research from before.

    The SAN we use is from a company called Tegile.

    How do I convince him not to do this? Or am I worried about nothing?

    • This topic was modified 5 years ago by  lmarkum.
  • You have the items separated in an AG to allow failover if there are issues. If he has an issue with the storage, then your failover won't work. No reason to introduce dependencies here.

     

  • My recommendation is just buy some more SSD.  Maybe even buy some less expensive spinning rust where you can move some filegroups to for things like audit tables and the like.  I don't see how deduping will help.  Perhaps page compression on large audit table and things like older parts of invoice detail tables will provide enough space.  The real key, though, is that data is going to continue to grow.  Buy the extra space because you're gonna need it.

    --Jeff Moden


    RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
    First step towards the paradigm shift of writing Set Based code:
    ________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

    Change is inevitable... Change for the better is not.


    Helpful Links:
    How to post code problems
    How to Post Performance Problems
    Create a Tally Function (fnTally)

  • Thanks Steve. That makes sense to me.

  • Thanks Jeff.  I appreciate it.  The databases are going to grow and dedupe isn't particularly effective for SQL.  I do have some databases with compression enabled where it makes sense.  Maybe I can re-evaluate those databases to see if there are columns I missed that might benefit.

    After this was posted the SysAdmin told me that the company was considering by a larger. all flash array from a different vendor, PureStorage.  I'll just need to make sure we account for these scenarios.

  • Thanks for the feedback.  IIRC correctly, PureStorage also makes it pretty easy to add on later.

    As a bit of a sidebar and especially with SSDs, I tend to "over-provision".  It's been worth every penny because there's never been a "holy crap! We're almost out of disk space!" moment and most people recommend over-provisioning SSDs for "life expectancy" purposes anyway.

    An example of my over-provisioning even enters my personal life.  After a decade of having an older laptop, I finally bought a new one.  It has 32GB of RAM (couldn't get the model I wanted with 64GB), 2TB of NVME SSDs, and 1TB of spinning rust for backups and speed comparisons.

    It cost me about 25% more than the standard offerings but I should be good for another decade or so and I don't have to worry so much about TempDB (which I frequently beat the hell out of) cutting a path in the carpet. 😀

    --Jeff Moden


    RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
    First step towards the paradigm shift of writing Set Based code:
    ________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

    Change is inevitable... Change for the better is not.


    Helpful Links:
    How to post code problems
    How to Post Performance Problems
    Create a Tally Function (fnTally)

Viewing 6 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic. Login to reply